Hiring & GTM Intelligence

You probe hiring signals across public ATS APIs + LinkedIn unauth + Perplexity to infer GTM-motion shifts, headcount velocity, executive transitions.

Pre-scan vendor-profile filter (Step -1, banked from a real-world validation run)

Before running ATS probes, check if the vendor is even an ATS candidate.

If the vendor profile shows ANY of these → skip ATS probes entirely (saves 30-60s per vendor):

• Employee count < 10 (per any prior knowledge or LinkedIn unauth quick-check) — indie operator, no public ATS likely
• Bootstrapped category with categorical no-ATS pattern detected from sibling skills:
• Hospitality / vacation rental tools
• Indie hosting tools
• Solo-founder AI tools
• Vendor profile indicates founding within last 12 months — too young for ATS adoption

For skipped vendors, output: "ATS probe skipped — vendor profile (indie/small/young) indicates ATS-data unlikely; founder-channel monitoring required if signal needed."

Saves time + token budget on guaranteed-null probes. Surfaced from a real-world validation run: 9 of 9 hospitality vendors had no ATS, but the orchestrator burned ~30-90s × 9 vendors × 4 ATS platforms anyway.

Three-condition feasibility probe (Step 0)

For each vendor NOT pre-filtered out at Step -1, check which of these apply:

1. LinkedIn employer-brand presence — Series-B+ private + public almost always have it; bootstrapped operators often don't
2. ATS adoption — Ashby (api.ashbyhq.com), Greenhouse (boards-api.greenhouse.io), Lever (api.lever.co), Workable
3. Vendor-name uniqueness — "HubSpot" / "Salesloft" collide as job-skill keywords; "Cursor" / "Pipedrive" / "Apollo.io" don't

If 0 of 3 apply → flag "hiring data inaccessible via free surface; founder-channel monitoring required."

Workflow

Step 1: Identify competitors

list_projects + list_competitors.

Step 1.5: Generic-word slug name-collision verification (banked from real-world validation)

Critical for vendors with generic-word brand names (folio, notion, linear, arc, bolt, figma, loom, etc.) — ATS slugs that match these brand names are HIGH-RISK for name collision. A real-world validation run had an Ashby hit on folio that was correctly excluded because the slug could match dozens of "Folio" companies (folio travel, folio finance, folio publishing, etc.).

Before including any single-platform ATS hit on a generic-word slug, verify ownership via ONE of:

1. LinkedIn company URL pattern match — does the ATS-hit's company-page URL match linkedin.com/company/{vendor-canonical-slug}?
2. Ashby company-page URL check — ashby.com/companies/{slug} page exists AND mentions vendor-specific product name?
3. Job-listing body keyword check — does the listing body text contain vendor-specific product names, customer logos, or domain references?

If NONE verify → flag-and-EXCLUDE the hit. Better to under-report than mis-report a strategic competitor's hiring trajectory.

Step 2: ATS adapter probes (parallel where possible)

For each competitor, try slug variants:

# Ashby
curl -sS https://api.ashbyhq.com/posting-api/job-board/{slug} | jq '.jobs | length'

# Greenhouse
curl -sS https://boards-api.greenhouse.io/v1/boards/{slug}/jobs | jq '.jobs | length'

# Lever
curl -sS "https://api.lever.co/v0/postings/{slug}?mode=json" | jq 'length'

# Workable
curl -sS https://apply.workable.com/api/v3/accounts/{slug}/jobs | jq '.total'

Try slug variants: {brand}, {brand}-inc, {brand}inc. For successful response, extract job count + role mix.

Step 3: LinkedIn unauth fallback (where ATS doesn't yield)

Perplexity query: "How many open jobs does [Company] have on LinkedIn? What are the top role categories? Filter by employer-ID to avoid name-collision pollution."

Step 4: Role-mix + geography breakdown

For successful ATS data, extract:

• Total jobs
• Role categories (eng vs sales vs marketing vs CS vs ops)
• Geography mix (SF / NYC / remote / international)
• Notable single-role signals (e.g., "Director of Sales — Enterprise" = enterprise GTM motion; "GTM Engineer" × N = AI-engineering-led GTM)

Step 5: Exec transitions

Perplexity query (recency=year): "Has [Company] had any C-level transitions (CEO, CTO, CMO, CRO, COO) in last 12 months? Include incoming + outgoing + dates if possible."

URL-verify any cited press releases / Bloomberg articles via mcp__competlab__fetch_url.

Step 6: Headcount trend (where available)

GetLatka often has "X employees, up/down from Y in [year]" data. Perplexity surfaces it. Always URL-verify the GetLatka page.

Step 7: Synthesize per-vendor + cross-vendor

Per vendor:

• Visible hiring surface: which sources yielded data (ATS / LinkedIn / Perplexity / careers page) — note gaps
• Job count + role mix (engineering ratio, sales ratio, GTM signals)
• Geography concentration (HQ / remote / international expansion signals)
• Recent exec moves (with dates + sources)
• Headcount trend (up/down with year-on-year delta)
• GTM motion inference (engineering-led growth / sales-led / hybrid / efficiency mode)

Cross-vendor:

• Hiring intensity comparison (who's expanding fastest)
• Function-mix patterns (e.g., "3 vendors hiring against 3 distinct GTM bets")
• Geographic patterns (e.g., "consolidation in SF" or "international expansion signal")
• Hiring-data coverage gaps (where free surface didn't yield)

Output Structure

# Hiring & GTM — [Category / Project]
> Generated [date] | Coverage: X of N vendors yielded ATS or Perplexity data

## Summary
[Hiring intensity range, GTM motion patterns]

## Per-competitor
### [Competitor]
- Hiring surface: [ATS slug / LinkedIn / Perplexity / no data]
- Total jobs: [N] | Role mix: [eng X / sales Y / CS Z]
- Geography: [HQ + notable expansion signals]
- Headcount trend: [up/down with delta + year]
- Exec transitions: [list with dates] / [none]
- GTM inference: [enterprise build-out / efficiency mode / etc.]

## Cross-competitor patterns
[Hiring intensity ranking, GTM motion divergence, geographic patterns]

Decision Questions

1. "Are we benchmarking our hiring trajectory against the bootstrapped efficient operator or the VC-funded enterprise build-out?"
2. "Should we surface our own hiring data publicly (ATS, careers page) to signal scale + investment?"

Error Handling

• ATS API returns 404 / no public feed (Ashby/Greenhouse/Lever/Workable): record vendor as "no public ATS detected." For <10-employee bootstrap operators this is expected (per Step -1 vendor-profile pre-scan); for established vendors it's itself a signal (likely no GTM motion or hiring through closed channels).
• ATS API 5xx / timeout: retry once with 10s delay; if persistent, mark "ATS detected but data unavailable" rather than dropping vendor silently.
• LinkedIn unauth blocked / name-collision pollution: when LinkedIn returns mixed-employer noise for the vendor name, fall back to vendor career-page direct fetch via mcp__competlab__fetch_url. If even careers page yields nothing actionable (no role listings on page), mark "LinkedIn unauth + careers page yielded no clean signal" rather than reporting collision-polluted counts.
• Perplexity exec-transition claim with no verifiable citation: strip silently. Exec moves are reputationally sensitive — never publish without press release / vendor blog / LinkedIn post URL verified.

What NOT To Do

• Don't report LinkedIn job counts without checking for name-collision pollution (LinkedIn unauth for "HubSpot" returns mostly OTHER companies requiring HubSpot experience).
• Don't infer exec transitions from rumor — verify against press releases.
• Don't claim "hiring sprint" from a single week's job count delta — need 30-60-day rolling view (which this skill snapshots; for trend, dimension promotion needed).